A lowly populated, transient β-sheet structure in monomeric Aβ1-42 identified by multinuclear NMR of chemical denaturation

Abstract

Chemical denaturation is a well-established approach for probing the equilibrium between folded and unfolded states of proteins. We demonstrate applicability of this method to the detection of a small population of a transiently folded structural element in a system that is often considered to be intrinsically fully disordered. The ¹H^N, ¹⁵N, ¹³C^α, and ¹³C' chemical shifts of Aβ^1-40 and Aβ^1-42 peptides and their M35-oxidized variants were monitored as a function of urea concentration and compared to analogous urea titrations of synthetic pentapeptides of homologous sequence. Fitting of the chemical shift titrations yields a 10 ± 1% population for a structured element at the C-terminus of Aβ^1-42 that folds with a cooperativity of m = 0.06 kcal/mol·M. The fit also yields the chemical shifts of the folded state and, using a database search, for Aβ^1-42 these shifts identified an antiparallel intramolecular β-sheet for residues I32-A42, linked by a type I' β-turn at G37 and G38. The structure is destabilized by oxidation of M35. Paramagnetic relaxation rates and two previously reported weak, medium-range NOE interactions are consistent with this transient β-sheet. Introduction of the requisite A42C mutation and tagging with MTSL resulted in a small stabilization of this β-sheet. Chemical shift analysis suggests a C-terminal β-sheet may be present in Aβ^1-40 too, but the turn type at G37 is not type I'. The approach to derive Transient Structure from chemical Denaturation by NMR (TSD-NMR), demonstrated here for Aβ peptides, provides a sensitive tool for identifying the presence of lowly populated, transiently ordered elements in proteins that are considered to be intrinsically disordered, and permits extraction of structural data for such elements.

Keywords: Chemical shift perturbation; Intrinsically disordered protein; Paramagnetic relaxation enhancement; Protein folding; Triple resonance NMR; Urea denaturation.

Figure 1.

Three simulated chemical denaturation curves for a typical folded protein (black), a protein that is highly resistant to chemical denaturation (red) and a transiently ordered IDP (blue). For this example, the chemical shifts of the folded and unfolded states are chosen to be 5 and 0 ppm, respectively, and are marked by dashed lines. Plots show the simulated chemical shifts under the condition of rapid exchange as a function of denaturant concentration, [D].

Figure 2.

Overlay of selected regions of projections of 34 700-MHz 3D HNCO spectra, recorded for four uniformly ¹⁵N/¹³C-enriched Aβ peptides at increasing urea concentrations, marked by decreasing intensity of color. Panels A and B correspond to the projections on the ¹⁵N-¹H^N plane; panels C and D are projections on the ¹³C’−¹H^N plane; panel E and F are projections on the ¹⁵N-¹³C’ plane for residues G37 and G38 (with the projection for panel E restricted to the 8.9–8.6 ppm region in the ¹H dimension). The black/grey dots mark the corresponding chemical shifts in the reference pentapeptide, with grey corresponding to the highest urea concentration. Urea concentrations (M) used for the titrations were Aβ^1–42: 0.00, 0.76, 1.63, 2.67, 4.04, 5.65, 7.15, 8.51; Aβ^1–42-Ox: 0.00, 0.87, 1.72, 2.62, 3.57, 4.86, 6.01, 6.98, 7.95; Aβ^1–40: 0.00, 0.81, 1.61, 2.70, 4.20, 5.78, 7.36, 8.85; Aβ^1–40-Ox: 0.00, 0.76, 1.58, 2.58, 3.45, 4.84, 5.95, 6.91, 7.78. Ref. peptide for G37 (QVGGQ): 0.00, 2.76, 5.38, 7.78; Ref. peptide for G38 (QGGVQ): 0.00, 2.72, 5.22, 7.82.

Figure 3.

Urea titrations of the backbone chemical shifts of the five Aβ peptides. Curves for other residues are shown in SI Figure S4. Dashed lines correspond to fits to Eq. 6, using global m and A values across residues I32-I41 (I32-V39 for Aβ^1–40). Legend colors match those in Figure 2.

Figure 4.

Contour plot of χ²/χ²_min, obtained for Aβ^1–42 from a 100 × 100 grid search of m and A, followed by best-fitting of the chemical shift parameters. The ‘+’ marker at A = 0.11 and m = 0.06 kcal/mol·M corresponds to the global minimum, χ² = 5.8 × 10⁻³. Analogous plots for the other four peptides are shown in SI Figure S5.

Figure 5.

Overlay of nine protein segments with secondary chemical shifts and residue types that best match the folded chemical shift (δ_f) values and residue types of I32-I41 obtained for Aβ^1–42 (cf Eq. 8). These segments resulted from a search over the SPARTA+ database that contains experimental chemical shifts for 580 proteins [50], with the best-matching segments listed in Table 3. Colors match those of the keys in Figure 8.

Figure 6.

Overlay of the 800-MHz ¹H-¹⁵N HSQC spectra of 75 μM Aβ^{1–42-A42C-MTSL} in 44 mM Tris-HCl, 10 mM NaCl, 2 mM EDTA, pH 7.4, 278 K, recorded before (black) and after (red) reduction of the MTSL spin label by addition of a 50-fold molar excess of sodium ascorbate, pH 7.4. Both HSQC spectra were processed with 15 Hz exponential line broadening in the ¹H dimension. Smaller panels correspond to expanded regions of the full spectrum.

Figure 7.

Peak intensity ratios in ¹H-¹⁵N HSQC spectra of the paramagnetic and diamagnetic forms of Aβ^{1–42-A42C-MTSL}. (A) Spectra recorded in the absence of urea (Figure 6). (B) Ratios obtained from spectra recorded under identical conditions except for the addition of 8 M urea. Red bars represent the upper limit for the attenuation ratios, based on the absence of detectable intensity in the paramagnetic state.

Figure 8.

Plot of the inverse sixth power of the distance between amide protons and the C^β atom of the residue that matches C42 in Aβ^{1–42-A42C-MTSL}. The intramolecular distances are shown for the ten best matching protein fragments listed in Table 3, nine of which contain an antiparallel β-sheet. For 2E7P, which contains a Gly residue at the position that matched C42, a pseudo-C^β atom was defined by extending the length of the C^α-H^α3 bond.